mljar / mljar-api-python

A simple python wrapper over MLJAR API.
https://docs.mljar.com/
Apache License 2.0
42 stars 10 forks source link

mljar.com app / new Experiment / Learning Algorithms is empty #8

Closed shadiakiki1986 closed 6 years ago

shadiakiki1986 commented 6 years ago

There used to be a list of checkboxes of algorithms to select from. Now it's empty.

pplonski commented 6 years ago

can you provide example how to reproduce this issue?

shadiakiki1986 commented 6 years ago

I think this is not an issue that is specific to mljar-api-python, but a general issue in mljar.com. If I click on "experiments" and "add new experiment", I get nothing under "use algorithm"

Screenshot

screenshot from 2018-01-22 10-32-36

pplonski commented 6 years ago

looks like you didnt set a task for a project, how did you created the project? with python-api or with web ui? can you provide example?

Thank you!

shadiakiki1986 commented 6 years ago

The project was created around a year ago maybe. It was created manually with mljar.com The day before running into this issue, I think the list of algorithms were showing up. But the same project is the one I'm using with the mljar-python-api, with example code same as in issue #6

shadiakiki1986 commented 6 years ago

So do you mean that this issue is specific to my own project only? If it is, I don't mind to just create a new project and work on it.

pplonski commented 6 years ago

Please try to create new project - if problem will occur please let me know. Thx!

shadiakiki1986 commented 6 years ago

Just created a new project manually, but when I run fit with mljar-python-api (with the same project name), a new project with the same name is created, and indeed it doesn't have a "task" set

shadiakiki1986 commented 6 years ago

I just realized that my target was continuous, whereas the original experiment was a binary classification. Reducing it to a binary class no longer creates a new experiment with the duplicated name.

shadiakiki1986 commented 6 years ago

As this was fixed with my paying attention that the target should be binary when the experiment is a binary classification, feel free to close this issue